Abstract: A method is provided. The method is implemented by a hyper-automation system to perform a process orchestration of a business process. The hyper-automation system includes a memory storing computer code for an autopilot component and for the process orchestration. The hyper-automation system includes a processor executing the computer code to cause the method to improve the process orchestration of the business process by automatically generating and adding implementation elements to the process orchestration using the autopilot component and natural language processing. The method includes receiving in prompts a natural language input. The method includes processing, by the autopilot component, the natural language input to determine integration locations for the implementation elements within the business process.

Abstract: According to one or more embodiments, a method is provided. The method is executed by a document understanding engine implemented as a computer program within a computing environment. The method includes receiving files by a document understanding project of the document understanding engine. The document understanding project includes a predefined domain set. The method includes automatically determining, by a generative artificial intelligence model of the document understanding engine, a domain from the predefined domain set for each of the files using content of each file. The method includes presenting, by the document understanding engine, the files documents according to the domain in a user interface of the document understanding engine.

Abstract: Systems and methods are provided for autonomously and semi-autonomously generating and executing automated test scripts from user stories using an AI-driven architecture. An AI agent receives user stories and associated acceptance criteria from one or more data sources. A reasoning engine processes the user stories to extract test scenarios and generates structured representations of the scenarios. An actuator module translates the scenarios into executable automated test scripts and autonomously integrates the scripts into continuous integration and delivery (CI/CD) pipelines. The actuator may further provision test environments, execute the automated test scripts, and collect telemetry, including execution results and system metrics. Optional embodiments include multi-agent configurations, replay mechanisms, escalation modules, and sandbox environments for semi-autonomous operation.

Abstract: A computer-implemented system and method for AI observability, distributed tracing, and privacy-preserving monitoring in AI-driven environments. The system assigns a unique trace identifier to AI-generated requests, enabling seamless tracking across multiple AI processing components. AI logs are enriched with structured metadata, including model execution parameters, inference confidence scores, and processing timestamps, to facilitate trace-to-log correlation. A federated trace management system ensures trace data is distributed across multiple storage locations while enforcing jurisdiction-based compliance regulations. To enhance privacy, the system applies privacy-preserving transformations, such as redaction, tokenization, and homomorphic encryption, preventing exposure of personally identifiable information (PII) in trace logs.

Abstract: A method is provided. The method is performed by a hyper-automation system performing agentic automations. The hyper-automation system includes a memory storing computer code for the agentic automations, for evaluation sets, and for evaluation score generation and enhancement. The hyper-automation system comprising a processor executing the computer code to cause the method to refine and improve over time the evaluation sets. The method includes determining whether an evaluation set of the evaluation sets for an agent is acceptable. The method includes running an evaluation set of the evaluation sets for an agent. The method includes generating an evaluation score based on the running of the evaluation set.

Abstract: A method of a hyper-automation system is provided. The hyper-automation system includes a memory storing computer code for one or more observer agents and one or more actor agents. The hyper-automation system includes at least one processor executing the computer code to implement and improve agentic automation through execution of the one or more observer agents and the one or more actor agents. The method includes monitoring the actor agents to determine an action of agentic automations for proposal or in performance. The method includes executing the observer agents utilizing context grounding to analyze the action by approving the proposal to enable the action to proceed or denying the proposal to prevent the action from proceeding or by validating or invalidating the performance of the action. The method includes completing the action by the actor agents subsequent to approval or validation from the analysis by the observer agents.

Abstract: A method is provided. The method is performed by a hyper-automation system performing agentic automations. The hyper-automation system including a memory storing computer code for the agentic automations, for evaluation sets, for agent health score generation and enhancement. The hyper-automation system including a processor executing the computer code to cause the method to refine and improve over time the evaluation sets. The method includes determining whether an evaluation set of the on evaluation sets for an agent is acceptable. The method includes, when the evaluation set is determined as acceptable, executing the evaluation set for the agent. The method includes, when the evaluation set is determined as not acceptable, enhancing the evaluation set for the agent by sufficiency determinations or health score assessments through iterative improvements until the evaluation set is acceptable.

Abstract: Systems and methods for analyzing, prioritizing, and potentially automatically generating robots implementing processes and/or process flows for robotic process automation (RPA) are disclosed. Artificial intelligence (AI) may be used to analyze business processes and/or process flows and look for possible candidates for automation or improvement of existing automations. Listeners (e.g., robots, separate software applications, operating system extensions, etc.) may be employed to listen in the background on user computing systems to mine data pertaining to workflow effectiveness and/or to identify new processes and/or process flows that may improve return on investment (ROI) for RPA.

Abstract: A method is provided. The method is implemented by a semantic automation builder executed on a processor. The method generating semantic automations of a robotic process automation. The method includes receiving user inputs corresponding to a target application, a written automation task, or steps to identify actions of the robotic process automation. The method also includes mapping each action to an activity to generate mapped activities, transforming the actions into the semantic automations based on the mapped activities, and providing the semantic automations in a user interface of the semantic automation builder to enable editing of the actions.

Abstract: Systems and methods for optimizing at least one of a process or an AI (artificial intelligence) agent are provided. The process and the AI agent are executed on one or more computing systems for performing a task. A visual representation of the execution of the process and the AI agent is generated. At least one of the process or the AI agent is optimized based on the visual representation. The optimized at least one of the process or the AI agent is output.

Abstract: A method is provided. The method is implemented by a hyper-automation system including processors executing computer code for agentic automations and an artificial intelligence trust layer and a memory storing the computer code. The method includes initially setting guardrails for an agent within the artificial intelligence trust layer of the hyper-automation system. The method includes operating the agent according to the guardrails to implement a workflow of the agentic automations including an action. The guardrails define a scope of the agent to control the operation of the agent as the agent implements action.

Abstract: A method is provided. The method is executed by a listener service implemented as a computer program executed by a processor. The listener service provides event listeners for change event management. The method includes initiating the event listeners specific to a target application and specifying change events for the event listeners that occur in the target application. The method also includes receiving, by the listener service from the event listeners, the change events in the target application, based on the specified change events, to capture the change events.

Abstract: An agentic memory method is provided. The agentic memory method is implemented by artificial intelligence (AI) agents. The agentic memory method includes storing vectors including context embeddings generated by advanced agentic extraction in response to queries in an agentic memory. The agentic memory method includes receiving and matching a complex query to the vectors of the agentic memory by an advanced agentic search to determine a context grounding. The agentic memory method includes providing the context grounding to an AI model with the complex query to enhance an accuracy of a response of the AI model and receiving the response to the complex query from the AI model.

Abstract: Systems and methods for extracting features from screen images for performing a task mining task are provided. A screen image depicting a user interface of a computing system is received. The screen image is preprocessed to generate a preprocessed screen image and processing results. Image features are extracted from the preprocessed screen image using a first machine learning based network. Text features and control element features are extracted from the processing results using a second machine learning based network. The text features and the control element features are encoded using a third machine learning based network to generate representative features of the screen image. A task is performed on the screen image based on one or more of the image features, the text features, the control element features, or the representative features. Results of the task are output.

Abstract: A method for translating an unstructured existing process into a well-structured process model is provided. The method is executed by a generation engine implemented as a computer program within a computing environment. The method includes analyzing elements or relationships of the unstructured existing process to detect model violations. The model violations include when the elements or the relationships fail at least one model requirement. The method include generating feedback suggestion types corresponding to the model violations and providing the feedback suggestion types in a user interface to automatically resolve the model violations.

Abstract: Systems and methods for automatically modifying a process by an AI (artificial intelligence) agent for adjusting user input of the process are provided. Process data associated with a process is received. The process is automatically modified based on the process data by the AI agent to adjust user input of the process. The modified process is output.

Abstract: Systems and methods for evaluating computer security issues of a computer system are provided. An alert for a computer security issue of a computer system is received. An initial AI (artificial intelligence) agent determines 1) a strategy for evaluating the computer security issue and 2) one or more additional AI agents from a pool of AI agents for evaluating the computer security issue. The computer security issue is evaluated according to the strategy using the one or more additional AI agents. Results of the evaluation of the computer security issue are output.

Abstract: Automation generation, execution validation, and healing by artificial intelligence (AI) agents is disclosed. In order to generate an automation, a user enters a description of the desired task to accomplish into a prompt or some other description and the AI agent attempts to perform the task by employing a screen understanding and document object model (DOM) extraction engine to extract screen content and/or using an object repository. A user may validate the steps that the AI agent will take when it executes before deploying the automation. AI agents may also be used to heal existing robotic process automations (RPA), application programming interface (API) automations, agentic orchestration processes (AOPs), etc. by searching the object repository for solutions for an RPA workflow and then also applying these solutions to other RPA workflows and then also applying the solutions to similar activities of other RPA workflows.

Abstract: Systems and methods for automatically requesting user input for performing a task by an AI (artificial intelligence) agent are provided. The AI agent is executed on a computing system for performing a task. It is determined, by the AI agent, whether to request user input during the performance of the task. In response to determining to request the user input, a particular user to request the user input from is identified by the AI agent. A request for the user input from the particular user is generated by the AI agent based on the task. The request for the user input is sent to the particular user. The performance of the task is continued by the AI agent based on the user input received from the particular user.

Abstract: A training/retraining lifecycle of artificial intelligence (AI)/machine learning (ML) models is managed using workflows with AI flows. Validation may be desired when an AI/ML model is called by an automation executing the workflow, such as an AI agent or robotic process automation (RPA) robot. This validation may include dynamic input from users. The automation may receive the dynamic input from the users and use this data for training a replacement AI/ML model or retraining the called AI/ML model. The state of the workflow may be preserved, both in training and serving. Workflows may be used to keep track of where the current execution is in the ML model lifecycle.